[time-nuts] How does NTP's local clock estimation work in detail?

Attila Kinali attila at kinali.ch
Wed Mar 11 16:33:16 EDT 2015

On Fri, 6 Mar 2015 17:23:14 +0100
Attila Kinali <attila at kinali.ch> wrote:

> Could someone point me to some papers that explain what NTP
> does locally? If possible in compact form.

I just send this of into a different direction, but i think it
is usefull for some people here as well:

> Did you find your answer yet? I would be interested too.

Yes, Magnus hint with the book[1] was very good. Its chapter 4
gives a nice overview of how it is done, but does not go too much
into the details. The other chapters cover a lot of the real world
issues that NTP in a nice and concentrated form.

For the details the book referes to [2] which covers the algorithm
for NTPv3 and [3] which covers NTPv4. There seem some typographical
problems with [3] though, which make me unsure about the formulas,
without really checking them, but [4] contains them without the
typographical problems. The book also mentiones [5], but i have
not had the time to have a look at it yet

[1] "Computer Network Time Synchronization", 2nd ed, by David L. Mills, 2010

[2] "Improved Algorithms for Synchronizing Computer Network Clocks",
by David Mills, 1995

[3] "Clock discipline algorithms for the network time protocol version 4",
by Mills, 1997

[4] "Adaptive Hybrid Clock Discipline Algorithm for the Network Time Protocol",
by Mills, 1998

[5] "An Algorithm to Synchronize the Time of a Computer to Universal Time",
by Levine, 1995


			Attila Kinali
< _av500_> phd is easy
< _av500_> getting dsl is hard

More information about the time-nuts mailing list